Research On Economic Operation Optimization Of Microgrid Based On Surrogate Model Particle Swarm Optimization Algorithm

The economic operation of the microgrid is an important factor in whether the microgrid can be widely promoted in the power system.However,because the microgrid combines different types of microsources,its optimization model presents the characteristics of non-convex,multi-constraint,etc.,which brings great difficulties to the economic optimization of microgrid.Aiming at the economic operation optimization problem of microgrid,this paper combines surrogate model technology,particle swarm optimization algorithm and machine learning technology to optimize the operation cost of microgrid,and analyzes the output results of each micro-source in microgrid.The main contents of this study are as follows:(1)This paper analyzes the different working modes of microgrid,and explains the importance and complexity of economic operation optimization of microgrid.Starting from the structure of microgrid,this paper briefly introduces the generation principle of micro-sources such as wind turbine and photovoltaic array,and explains that each distributed micro-source is easily affected by many factors such as weather and user demand,so as to illustrate the complexity of economic operation optimization of microgrid.This paper introduces Kriging model and RBF model,and explains their selected performance indicators in terms of precision and dimension.It also discusses the advantages of choosing Kriging model and RBF model for surrogate ensemble in order to solve the "disaster of small data".(2)Aiming at the shortcoming that the swarm intelligence optimization algorithm is difficult to optimize the economic operation of the microgrid,a surrogate-assisted evolution algorithm based on a single improved Kriging model is proposed.The algorithm dynamically improves the correlation function matrix of the Kriging model.On this basis,it combines the global search with the local search.By combining the relatively simple surrogate model with PSO,the quality of solution is guaranteed and the convergence speed is improved.The proposed algorithm(Im KPSO)is tested with 13 test functions in CEC2006 and 9 test functions in CEC2017,and the effectiveness of Im KPSO is verified.Finally,taking the economic optimization operation of microgrid as an example,the simulation results show that Im KPSO has strong advantages in solving the problem of economic optimization operation of microgrid.(3)Aiming at the shortcoming of insufficient accuracy of single model surrogate-assisted evolution algorithm in the optimization of microgrid economic operation,a surrogate-assisted evolution algorithm based on ensemble learning is proposed.The algorithm trains multiple surrogate models on the same initial sample set,including Kriging model and RBF model.Finally,these multiple surrogate models are combined into a surrogate ensemble.Compared with the single surrogate model,the accuracy of the surrogate ensemble is greatly improved.By testing the various components of ELe PSO on the benchmark function,the effectiveness of the algorithm is verified.The comparison of several latest surrogate-assisted evolutionary algorithms with ELe PSO shows that the proposed algorithm has good performance in dealing with complex optimization problems.At the same time,the economic optimization operation of the microgrid is taken as an example for simulation verification.The experimental results show that compared with Im KPSO,ELe PSO has more obvious advantages in solving the economic optimization operation of the microgrid.